Hi Bill,

Bill.Holler at Sun.COM wrote:

> Hi Aubrey,
> 
> I was only able to look through the first half which looks good.  :-)

Thanks for the comments, ;-)

> 
> If I remember correctly the ACPI spec defines a latency > 1000 to mean
> that c-state is not supported?  If so should cpu_acpi_cache_set()
> check for cstate->cs_latency > 1000 in its for loop?
> 

That might not be true. Evaluating _CST objects returns how many C
states supported,
and the latency as well. Different C-state has different latency, it
doesn't determine deeper c-state supported or not.
The following is some info from my broken machine.

C type          Latency         Power consumption
C1              1 - us          1000 - mW
C2              1 - us          500 - mW
C3              162-us          100 - mW

Thanks,
-Aubrey

Reply via email to